1. Introduction

Optical tweezers are a technique to trap and manipulate micro- and nano-scale materials using optical force caused by a tightly focused laser beam [1–4], which becomes essential tools in advancing the forefront of nanoscience and life science, particularly in biological application. Another technique, optothermal manipulation, utilizes lasers to create a temperature gradient for controlling the mechanical motion of particles along the gradient [5–8]. For instance, an early report demonstrated that fluorescent-labeled DNA molecules diffuse from hot to cold areas, or vice versa, depending on whether the solution chamber is at room temperature or 3 °C, respectively [9]. Several driving forces, including thermophoresis, thermo-electrophoresis, thermo-osmotic slip flows, convection flow, and Marangoni flow, play roles in manipulating the particles and are complexly interrelated within a single system. These forces are sensitive to various environmental parameters including solution temperature [9], particle surface charge [10], thickness of the bulk solution [11], substrate surface [12–14] and concentration of molecules dissolved in the solvent [15].

With the recent interests of plasmonic optics, locally enhanced electric fields yielded on the plasmonic nanostructures have been utilized for efficient opto-thermal trapping [16–23]. For instance, thermo-plasmonic substrate (a porous Au film) served as an absorber for visible laser, and individual Au and Ag nanospheres were trapped at the laser focus by controlling the environmental ionic distribution [14]. In addition, high-density assembly of bacteria with a high survival rate were demonstrated using honey-comb-like photothermal film with 1064 nm laser [20]. Notably, all these experiments depend on intermediate materials to indirectly convey laser energy to the solvent via metallic nanostructures.

The water solution can be directly heated by exciting its vibrational modes with infrared lasers, without the use of intermediate materials. Water exhibits absorption originating from the OH vibrational modes at approximately 1.5, 2 and 3 µm wavelengths as represented in Fig. S1 of Supplementary Information 1 [24,25]. Overtone modes are excited by the former two wavelengths, while fundamental modes are excited by the light of 3 µm wavelength. Therefore, the extinction coefficient at 3 µm wavelength is significantly high. For instance, that at 2 µm and 3 µm are roughly 4 and 1050 times larger than that at 1.5 µm. In early opto-thermophoresis experiments conducted by D. Braun group, lasers ranging from 1.4 to 1.5 µm were typically used to heat water by exciting the OH vibration modes, which reveals the fundamental dynamics of these studies [9,26–28]. To enhance the water absorption, we previously utilized a 2 µm Tm-doped fiber laser for directly heat the water solvent without using any metallic nanostructures [29].

In this article, we employ a 3 µm laser, which has significantly high absorbance due to the fundamental mode of the OH vibration. To attain a laser wavelength of 3 µm, we develop an Er:ZBLAN fiber laser with an external grating cavity to finely tune the wavelength of continuous-wave laser. The tuning range, from 2700 to 2826 nm, covers from the onset of absorption toward the peak, as illustrated in Fig. 1(h). Hereafter, we represent our results of opto-thermal manipulation with the Er:ZBLAN fiber laser, which accelerates the assembling speed much faster than other laser wavelengths by simply irradiating the laser into the water solvent without preparing any absorber structures and films.

 

Fig. 1. (a) Setup of 3 µm continuous wave Er:ZBLAN fiber laser with external cavity. (b) Setup for opto-thermophoresis with Er:ZBLAN fiber laser system. Inset shows the configuration of the focused laser beam and sample of particles in water. Abbreviations used: DM = dichroic mirror; BM = Beam splitter; OL = Objective lens. (c-g) Beam profile of the laser after the Ge plate. Scale of white bars is 5 mm. (h) Spectra of 3 µm Er:ZBLAN fiber laser and absorbance of water measusred by ATR-FTIR spectroscopy. ATR correction is applied to the water spectrum.

Download Full Size | PDF

2. Optical setup and samples

2.1 Laser generation of 3 µm mid-infrared Er:ZBLAN fiber laser

The schematic diagram of the tunable ErF₃-doped ZBLAN fiber laser is shown in Fig. 1(a). The optical setup is primarily based on the approach described in the literature [30]. The pumping source is a fiber-coupled laser diode, which has a numerical aperture (NA) of 0.22 and delivers an output power reaching up to 10 W at a 976 nm wavelength. The gain medium is a 3-meter-long ErF₃-doped double clad ZBLAN fiber (provided by Fiber Labs Inc.), with a core diameter of 32 µm and the NA of 0.15. The dopant concentration of Er ions is 6 mol.%. The fiber is wrapped by a circular-shaped inner-cladding of 300 µm diameter and 0.52 NA. The pump radiation is collimated by a high precision CNC-Polished Aspheric Lens (AR Coated to 650 - 1050 nm), with 20 mm focal length. The collimated beam is focused into the gain fiber inner-cladding by using a Zinc Selenide (ZnSe) Plano-convex lens (with AR Coated to 2 - 13 µm), with 50 mm focal length. To avoid the broadband feedback caused by uncoated fiber facets, the ZBLAN fiber is 0◦ polished at the pumping launch end, and 8◦ angled-polished at the output end. The output laser from the ZBLAN fiber is collimated by the ZnSe Plano-Convex Lens, (AR Coated to 2 - 13 µm), with 25.4 mm focal length. The collimated beam is reflected by a grating mounted on a rotation stage. Mid-IR ruled reflective diffraction grating, with 3.1 µm central wavelength, and the dimension of 12.5 × 25.0 × 9.5 mm is used to tune the wavelength of the output. For the trapping experiment, we tune the laser wavelength to 2700 nm (3704 cm^–1), 2722 nm (3674 cm^–1), 2760 nm (3623 cm^–1), 2790 nm (3584 cm^–1) and 2826 nm (3539 cm^–1) by adjusting the grating angle with the rotation stage (see Fig. 1(h)). The laser output power is stabilized at these wavelengths where the ro-vibrational absorption of the ambient vapor is lower. The mid-infrared laser output is extracted by a rectangular dichroic mirror, with high-transmission at 976 nm and high-reflection at 2.6-2.8 µm at 45-degree incident angle. In addition, a Ge plate is added to the optical path in order to cut the remaining pumping laser. An infrared (IR) camera (Spiricon, Pyrocam IIIHR) is used for observing the beam profile (see Fig. 1(c-g)). The beam diameter after the Ge plate is approximately 6 to 10 mm (D4σ) depending on the laser wavelength, which fulfills the back aperture of the objective lens. The spectrum of the output laser is detected by an Optical Spectrum Analyzers (Thorlabs, OSA207C).

2.2 Laser microscope and samples

The generated 3 µm laser is introduced to a home-built microscope for opto-thermal trapping experiment as shown in Fig. 1(b). A maximum output power is 622 mW (at 2774 nm) and the beam diameter fulfill the back aperture (5.1 mm in diameter) of reflective type objective lens (Thorlabs, NA 0.5). The collimated laser is reflected up by the dichroic mirror and focused at the bottom-glass/solution interface as illustrated in the inset of Fig. 1(b). The laser power is measured after the objective lens and its transmittance is approximately 16%. During the trapping, the images are recorded by the camera with standard transmission images with Kohler illumination. To measure the elevated temperature at the focal point, we monitor the changes in fluorescent intensity of Rhodamine B during 3 µm laser irradiation. The dyes are dissolved in water, and a 488 nm continuous wave laser (Omicron Laserage, LuxX 488-100) is used for excitation. The excitation laser power at the focus is 0.1 mW. The molar concentration of the Rhodamine B is 88µM. A fiber core diameter guiding the signal to spectrometer is 100 µm, therefore a confocal effective detection area is approximately 2 µm concerning the magnification (x50) of our microscope. For target samples, we used polystyrene particles (1 µm in diameter). The mother solution is diluted by sodium chloride aqueous solution (50 µg/mL, 0.86 mM) to the appropriate particle concentration in order to trap the particles at the laser focus. The sample solution is sandwiched by two glass substrates (Matsunami glass) with a both-side tape spacer of 120 µm thickness. Additionally, the particle trajectories were determined using the method we previously employed [29].

3. Results

3.1 Comparison between 1956nm and 2700 nm lasers

First of all, to highlight the differences from our previous results obtained with the 1956nm laser [29], we demonstrate the opto-thermal trapping of the 1 µm polystyrene particles employing the 2700 nm laser, with identical laser power of 2.7 mW. Initially, the particles in water are dispersed homogeneously and undergo Brownian motion at the room temperature (approximately 25°C). Upon the laser irradiation, the temperature rises due to the absorption of the lasers by the water, which is based on O-H vibrational modes, thereby creating a temperature gradient around the laser focus. With the 1956nm laser, the particles are gradually trapped at the focus, subsequently forming the assembly as shown in Fig. 2(a). Even when the laser is irradiated for 5 s, the particles do not migrate at the focus.

 

Fig. 2. Transmission images of 1 µm polystyrene particles trapped and assembled by the lasers of (a) 1956nm and (b) 2700 nm, respectively. The laser power at the focus is approximately 2.7 mW. The images show at 1, 3, 5, 10 s after turning on the laser. A background color for (a) is different from others because a different dichroic mirror is used to reflect the 1956nm laser into the microscope. The length of the black bars is 20 µm.

Download Full Size | PDF

When the laser wavelength is altered to 2700 nm, the particle assembly formation is indeed accelerated, as observed in Fig. 2(b) (refer also to the Visualization 1 for a comparison between the results of the 1956nm and 2700 nm lasers). The particles convey toward to heated focus along the glass substrate from far fields, which is possibly due to the thermal convection. After the accumulation, a close-packed hexagonal structure forms around the focus, and its resultant assembly diameter is approximately 20 µm after 10 s irradiation. Interestingly, each particle constituting the assembly is fixed and stable, not exhibiting random motion; however minor fluctuations of the particles are observed in the assembly prepared by the 1956nm laser (see 10 s at Fig. 2(a)). Consequently, the 2700 nm laser can accumulate the particles more rapidly and tightly compared to the 1956nm laser.

When we only used deionized water as the solvent, the particles convey toward to the heated focus, however the particles were not immobilized at the focus and were pushed along the light propagating direction (the Visualization 2). Additionally, the trapping and assembly formation were not observed when the water solution is replaced to heavy water (with the NaCl addition) which has a red-shifted absorption vibrational band compared to water. The thermal diffusion is also much suppressed (Supplementary Information 2 and the Visualization 3). This fact suggests that the driving force in our system is indeed the opto-thermal effects, instead of the optical trapping force.

To attain the assembly of similar size using the 1956nm laser, a laser power of approximately 7.6 mW is required (see Fig. S3 in Supplementary Information 3). The photon density of 1956nm laser (7.6 mW) is 425 µW/µm², and that of 2700 nm laser (2.7 mW) is 79 µW/µm², respectively. In these calculations, we assumed that the focal diameters to be the same as airy disc sizes when the NA of focusing objective lens is 0.5. Concretely, the focal diameters of 1956nm and 2700 nm are estimated to be 4.77 and 6.59 µm, respectively. From the estimation, it indicates that the use of the 2700 nm laser requires approximately 5.4 times lower photon density. From the viewpoint of an extinction coefficient of water [24,25], that at 2700 nm (1.9 × 10⁻²) is approximately 21.5 times larger than that at 1956nm (8.83 × 10⁻⁴). Hence, the experimental enhancement (5.4 times) is lower than the simply estimated enhancement (21.5 times). From this moderate discrepancy (approximately four times difference), we suggest that the observed behavior cannot be simply explained by the magnitude of the extinction coefficient as we explain details in discussion section.

3.2 Higher laser power of 2700 nm laser

As the power of the 2700 nm laser increases, the particle trapping is much accelerated; however, a hollow-like area emerges at the center of the assembly, as presented in Fig. 3. In the initial stage of trapping (see Fig. 3(b)), the particles are not exactly trapped at the center but are localized surrounding the focus. Several particles are assembled forming a cluster nearby the surrounding area. Subsequently, this cluster is trapped to the focal center, and the assembly evolves as depicted in Fig. 3(c-e) and Fig. 3(g-i). Upon close inspection, the particles within the focus exhibit a different appearance compared to the peripheral particles. In particular, boundaries of each particle are not as distinct as those of the peripheral ones (see Fig. 3(f)), implying that the particles may partially merge with each other due to laser heating (see detail for Supplementary Information 4). Besides, the assembly diameter is 30 µm, and the number of the trapped particles is approximately eight hundred particles within 7 s. In addition, the particles are started to be trapped at the second layer of the assembly. Interestingly, the second layer is not formed at the focus even for a long exposure time. Although the hollow-like area is created at the focus, vast amounts of particles can be trapped at the peripheral region of the laser focus by just irradiating the 2700 nm laser to water solution.

 

Fig. 3. Transmission images of 1 µm polystyrene particles trapped and assembled by the 2700 nm laser of 7.2 mW. (a-e) The assembling dynamics in the early time after turning on the laser. The focal spot is marked with cross-dashed lines. The length of the black bars is 10 µm. (f) shows the enlarged view of (e), where the particles existing at the center. (g-i) The images show at 1, 5, and 7 s, respectively. The length of the black bars is 20 µm.

Download Full Size | PDF

3.3 Laser wavelength dependence

As the laser wavelength approaches nearer to the OH vibration absorption peak, the assembly either reduces in size or maintains a size comparable to that formed by the 2700 nm laser, as shown in Fig. 4. Additionally, the hollow area becomes apparent, especially when the laser wavelength nears the absorption peak. For example, a small hollow barely emerges at the center with a wavelength of 2722 nm (see Fig. 4(c)). Furthermore, no particles are present at the focus for the wavelengths of 2760, 2790 and 2826 nm, even though the particles were present for the 2700 nm laser of higher power, as we mentioned previously (see Fig. 3). These observations imply that a repulsive force increases with longer wavelengths. Another feature is that the assembly only expands to a limited size (approximately smaller than 20 µm) when employing the 2790 nm and 2826 nm lasers. Once it reaches this saturated size, the particles located at the assembly edge are circulated by the convection flow (see the Visualization 4, especially after 20 s). In addition, the temperature at the focus is estimated by using the thermal sensitive fluorescent dyes as we explained the detail in Supplementary Information 5. Briefly, the elevated temperatures for 2700, 2722, 2760, 2790 and 2826 nm were 72.5 ± 3.1, 68.5 ± 3.5, 64.9 ± 2.9, 62.9 ± 3.6 and 63.5 ± 4.5 °C, respectively, which is comparable across all the wavelengths. Besides, when the laser power exceeds about 10 mW, the bubbles are generated, inhibiting the opto-thermal trapping.

 

Fig. 4. Transmission images of 1 µm polystyrene particles trapped and assembled by the lasers of (a) 1956nm (b) 2700 nm (c) 2722 nm (d) 2760 nm (e) 2790 nm, and (f) 2826 nm, respectively after 10 s laser irradiation. The laser power at the focus is approximately 2.7 mW. The length of the black bars is 20 µm.

Download Full Size | PDF

3.4 Single-particle tracking analysis

To gain deeper understanding of assembling behaviors, we performed a single-particle trapping experiment. The coordinates of particle being pulled toward the focal center are extracted by single-particle tracking analysis, as previously utilized [29]. In details, trapping stiffnesses in x, y focal plane are calculated by using the equation of ${k_x} = {\textrm{k}_\textrm{B}}\textrm{T}/{x^2}$, and ${k_y} = {\textrm{k}_\textrm{B}}\textrm{T}/{y^2}$, where ${x^2}$ and ${y^2}$ are position variance in x- and y- directions. This estimation is based on the assumption that the particle located near to the focus follows the Hooke’s law, as it is typically used for optical trapping and opto-thermal trapping experiments to evaluate the trapping strength [31,32]. Following this, we divided the results into two sections. First, we illustrate the trajectory of the particles as it approaches the center from outside. Next, we discuss the spatial distribution of the particle after it is trapped.

3.4.1 When the particle approaches to the focus

Figure 5(a) displays representative trajectories of single particle being pulled toward the focus. Initially, the individual particles are located at the top-right of the field of view before the laser irradiation, approximately 40 to 50 µm from the focus. As illustrated by the blue trajectories in Fig. 5(a, b), the particle reaches the focus more rapidly with the 2700 nm laser than with other wavelengths. As indicated in the blue curve of Fig. 5(c), the particle velocity within 0 to 16 s time frame is larger compared to other wavelengths. Even though the extinction coefficient is the smallest among the five wavelengths, the particle arrives first with the 2700 nm laser, which correlates with the observation of larger assembly size observed above in Fig. 4(b). As we explain later in discussion section, we suppose that the convection flow is the key element in understanding this phenomenon. The particle trapped with the 2722 nm laser, illustrated by red trajectories in Fig. 5(a, b), arrives second, while the arrival times for the remaining three wavelengths (2760 to 2826 nm) are nearly the same.

 

Fig. 5. (a) Trajectories of the single particles trapped by the 2700, 2722, 2760, 2790, 2826 nm lasers, respectively, within 16 s. The origin is the place of laser focus, and the particle initially located at the top-right of the laser focus. (b, c) The x-position and velocity of the particles as a function of the irradiation time. The vertical dotted lines indicate the time when the particle is trapped at the focus. The arrows denote the maximum velocity for each wavelength. The time interval for calculating the velocity is 0.67 s (corresponds to 20 frames).

Download Full Size | PDF

Here we focus on the particle’s instantaneous velocity just before it becomes trapped. As a result, the instantaneous velocity tends to increase as the laser wavelength approaches the peak of the O-H vibrational modes. Concretely, the velocities are 7.4 µm/s, 10.2 µm/s, 8.2 µm/s, 12.4 µm/s, 13.3 µm/s for 2700 nm, 2722 nm, 2760 nm, 2790 nm, 2826 nm, respectively. The time interval for these velocity calculations is 0.67 s which corresponds to 20 frames. When the time interval is reduced to a single frame (33.3 ms), these values become, 27.8 µm/s, 37.2 µm/s, 33.7 µm/s, 38.0 µm/s, and 48.8 µm/s for 2700 nm, 2722 nm, 2760 nm, 2790 nm, and 2826 nm, respectively.

3.4.2 Particle distribution nearby the focus

Once trapped at the laser focus, the particle localizes in the surrounding region of the focus, as illustrated in Fig. 6(a). The resultant distribution profile is circular for all wavelengths. The particle cannot localize at the center due to a hollow area that presents a potential barrier, repelling the particle away from the center. Notably, this hollow area can be filled by subsequent particles when the laser wavelength is 2700 nm. On the other hand, the hollow area remains unfilled at other four wavelengths, as we explained in Fig. 4(c-f).

 

Fig. 6. (a) The particle distribution around the laser focus for each laser wavelength. (b) Histogram of the particles trapped nearby the focus for each laser wavelength. The laser wavelengths are (i)2700, (ii)2722, (iii)2760, (iv)2790, and (v)2826 nm.

Download Full Size | PDF

To estimate the diameter of circle distribution and optical trapping stiffness, Fig. 6(b) represents a probability density across the radial direction. One aspect is that the diameter of the distribution slightly expands with elongating the laser wavelength as summarized in Table S1 in Supplementary Information 6. We suggest that this expansion of the radius is attributed to the thermal repulsive force of the particle heating. Although the particle distribution is not a harmonic oscillation centering the origin (focus), we estimated the trapping stiffness from this distribution across the radius. As summarized in Table S1, the trapping stiffness per laser power increases as elongating the laser wavelength, while the stiffness at 2826 nm wavelength slightly drops. To show the best performance, the trapping stiffness per power is approximately 15 times higher comparing the results of the 2790 nm and 1956nm laser previously we used.

4. Discussion on particle assembly mechanism

Based on all the above experimental observations, we here extrapolate the assembling mechanism in terms of thermophoresis, convection flow, thermo-osmotic slip flows and a three-dimensional elevation of water temperature within the focal cone, as illustrated in Fig. 7. Upon laser irradiation, the 3 µm mid-infrared lasers resonantly excite the water's vibrational modes, generating a temperature gradient. It should be pointed out that, the three-dimensional heat profiles within the focal cone are different for each wavelength, due to the different extinction coefficients. In particular, as illustrated in Fig. 7(a, b), the heat profile becomes a cone-like (volume-heating) and a point-like (point-heating) for the 2700 nm laser and 2826 nm laser, respectively. In detail, the solution thickness is approximately 120 µm, and the penetration depth estimated from the extinction coefficients [25] are 22.6 and 2.97 µm, respectively.

 

Fig. 7. (a) Illustration for explaining the assembling dynamics. (a) Cone-like heat profile is generated by the 2700 nm laser irradiation. (b) Point heat profile is generated by the 2826 nm laser irradiation. Blue, green, violet and red arrows indicate thermo-osmotic slip flow, thermal convection flow, vertical thermophoretic force, and heat repulsive force, respectively.

Download Full Size | PDF

In the case of volume-heating, the convection flow is amplified compared to the point-heating. The water density decreases within the focal cone, causing the water to rise upward and creating the enhanced convection flow denoting by the green arrows in Fig. 7(a). This thermal convection conveys the particles from far fields toward the heated focus. When the particles approach on the glass substrate, the thermo-osmotic slip flows drive the particles along the interface toward the heated spot (see blue arrows). As the experimental fact, all the particles are approaching to the focus with sliding on the glass substrate as we can observe in the Visualization 1 and 4 (the appearance of the particles are black in the transmission image representing they are on the glass substrate).

The addition of NaCl is essential for trapping and assembling particles at the focus. In the deionized water, the particles similarly convey toward the focus but they are not trapped at the focus. We suggest that the NaCl addition triggers the thermo-osmotic slip flow, as it brings the particles closer to the substrate by neutralizing their surface charges. Then, the resulting interfacial flow effectively assembles the particles at the heated spot. This mechanism appears consistent with prior findings where gold nanoparticles were trapped by the thermo-osmotic slip flows. These flows are triggered by modulating van der Waals and double layer interactions at a gold-liquid interface with the addition of NaCl [14]. Subsequently, a vertical thermophoresis [12] plays a role for stabilizing the formed assembly (see a violet arrow). Since, the glass substrate has higher thermal conductivity than the water, thermal gradient points vertically toward to the glass substrate, resulting in the confinement force stabilizing the formed assembly.

The formation of a hollow area at the center of the assembly can be attributed to the heat elevation of the particles themselves (see red arrows). Although the laser wavelength is not at the resonance peak of C-H vibrational modes (around 3.3 µm in wavelength), we suppose that some light absorption still occurs, resulting in the heating of the particles. In support of this, we observed aggregates formed by several polystyrene particles adhering to the glass substrate after the laser was turned off. Notably, the glass transition temperature of polystyrene is approximately 100 °C, aligning closely with the temperature elevation measured using fluorescent dyes (the estimated temperature was 60 to 70 °C). Additionally, the heated and softened polystyrene particles may be merged by these driving forces, resulting in the formation of aggregates that adhere to the glass substrate.

For the point-heating with 2826 nm laser, the convection flow becomes moderate (see green arrows) compared to the volume-heating. On the other hand, the thermo-osmosis slip flow may be enhanced due to higher extinction coefficient (see blue arrows). In the experiment, the particle exhibits an increased instantaneous velocity and experiences tighter confinement, suggesting a steeper local temperature gradient for longer wavelengths. Therefore, it is understandable that both the number of the particle being assembled becomes less and assembling speed become slower. All in all, we believe that these multiplex driving forces and factors (thermal convection, slip flows, NaCl addition, vertical thermophoresis, heated particles, and volume and point heating) play a dominant role for the assembling in our system. For the comprehensive and quantitative understanding, we plan to conduct a hydrodynamic simulation in the near future.

In relation to volume heating, there is a literature presenting the effect of volume heating on a bubble generation that leads to assembly formation [17]. Specifically, microbeads covered with a vast number of Ag nanoparticles were assembled around the bubble generated at a focus of near-infrared laser due to the photothermal effect. The bubble emerged at the focal spot as these beads absorbed a near-infrared laser through localized surface plasmon excitation. The study also theoretically calculated the bubble size, considering the effect of the heat generated from the focal cone area, suggesting that the volume heating contributes to the bubble generation. Compared to this, within our system, the bubble is not generated and the volume heating only contributes to the convection flow. In particular, the density of water in the focal cone is reduced, thereby initiating the stronger convection flow with the 2700 nm laser compared to the 2826 nm laser.

On the other hand, it is important to address the effect of horizontal thermophoresis and thermoelectric effects relying on the thermally induced charge separation. Firstly, when the particles are away from the substrate, the horizontal thermophoresis dominates the thermo-osmotic slip. In fact, we observed that the particles displaced from the substrate are repulsed away from the focus when the laser is on (see Visualization 1 and 4). Secondary, negatively charged particles cannot be trapped at the heated spot in the NaCl solution concerning the ion redistribution [33,34]. Therefore, we suppose that these are not the primary factor for the assembling in our system. Besides, an unrevealed issue is that the estimated temperatures are similar or even smaller for longer wavelengths; this may be clarified by examining the three-dimensional heat profile with laser scanning confocal microscope in future studies.

5. Conclusion

We have successfully demonstrated an optothermal manipulation with a 3 µm Er:ZBLAN fiber laser which can excite OH fundamental vibrational modes and directly heat the water solvent, thereby yielding a temperature gradient to trap and assemble particles at the laser focus. A home-built ZBLAN fiber laser with an external grating cavity can selectively emits a continuous-wave laser in spectral region ranging from 2700 nm to 2826 nm in wavelength. We observed that an assembling speed of the particles become indeed faster when using the 2700 nm laser, compared to both the longer wavelengths and the 1956nm laser used in our previous experiment [29]. Even though, a hollow area appears at the assembly center with the higher laser power, we can accumulate the particles much faster compared to the use of 1956nm laser. On the other hand, an instantaneous velocity of the particle just before it is trapped becomes fastest with the 2826 nm laser, wherein the water absorbance is significantly higher at this wavelength. We propose that the dynamics underlying the formation of particle assembly can be understood through considering the interplay of thermal convection, thermo-osmotic slip flows, thermophoretic force, and the target particle heating, with the balance among these factors altering based on volume or point heating. The originality of the present method lies in the use of 3 µm mid-infrared laser that can directly heat the water solution by exciting their vibrational modes, whereas most recent studies on plasmon-based optothermal trapping involve indirect heating of water through plasmonic nanoparticles and nanostructure with visible and near-infrared lasers [16–23]. With our method, we only need to irradiate the 3 µm mid-infrared laser into the water solution without any additional metallic nanoparticles or nanostructure preparations.